CN104503236B - A kind of decomposition furnace outlet temperature sliding-mode control based on regression model - Google Patents

Abstract

The invention discloses a kind of decomposition furnace outlet temperature sliding-mode control based on regression model, first according to cement predecomposition technological process and site operation personnel's experience, choose the input variable for feeding coal amount as model, according to on-site actual situations, it is typical condition to select 840 DEG C~860 DEG C of decomposition furnace outlet temperature；Then according to historical data, the decomposition furnace outlet temperature mathematical modeling based on regression analysis is set up；Optimum control amount is finally asked for using adaptive tendency rate, adaptive sliding mode controller is set up, with stronger robustness and consistency.The present invention can accurately realize the control of decomposition furnace outlet temperature, to realize that the optimal control of industry spot dore furnace provides new approaches.Adaptive sliding mode controller is set up, with stronger robustness and consistency, the control of decomposition furnace outlet temperature can be accurately realized, to realize that the optimal control of industry spot dore furnace provides new approaches.

Description

A kind of decomposition furnace outlet temperature sliding-mode control based on regression model

Technical field

The invention belongs to technical field of automation in industry, more particularly to a kind of decomposition furnace outlet temperature based on regression model Sliding-mode control.

Background technology

China's cement total output is ranked the first in the world, and annual production is up to more than 20 hundred million tons, accounts for more than the 50% of global total amount, because And realize cement industry energy-saving research emphasis and focus as current manufacture of cement using automatic technology.Manufacture of cement Predecomposition process be one of core link of manufacture of cement, its coal consumption is big, accounts for whole cement production process coal consumption total amount 60%.Therefore, optimal control is implemented to the link and energy-saving be significant is realized to cement production enterprise.

As the nucleus equipment of predecomposition technology, dore furnace be responsible for burning heavy in precalcining system, heat transfer and The task that material is decomposed.Because the working condition change of raw material predecomposition process is frequent and measuring control point is few, this causes in reality Often occurs the phenomenon of calciner temperature fluctuation in production.Temperature is too high easily to cause preheater skinning, influences kiln system Normal operation；Temperature is too low, then causes too low into kiln resolution ratio, increases kiln system burden, it is impossible to give full play to the work of dore furnace With.Therefore, the control of decomposition furnace outlet temperature, both cement production enterprise is realized it is energy-saving have great importance, also contribute to Manufacture of cement is normally carried out.

In order to realize the control of decomposition furnace outlet temperature, it is very heavy to set up suitable decomposition furnace outlet temperature mathematical modeling Want.Document (takes the numerical modeling powder technologies of De-Nol cement predecomposition calcination process dore furnaces, 2007,117 (1)：81- 85.) from kinetics, the dynamics mathematical model of coal dust firing and carbonate decomposition is set up.Influence is not investigated Relation between the principal element and calciner temperature of calciner temperature time-varying.(Yao Weimeng, face zhang person of outstanding talent, Zhu Jing cement is returned document Fuzzy control [J] instrument and meter for automations of rotary kiln calciner temperature, 2001,22 (2)：35-37.) propose a kind of calciner temperature Fuzzy control method.Zhang offer (application [J] industrial instrument of Wang Haifeng, Zhu Jing the Fuzzy Predictive Controls in manufacture of cement with Automation equipment, 2003,02：A kind of fuzzy prediction control algorithm of improved calciner temperature 10-13.) is given, and is provided Its operational effect comparative analysis controlled with traditional fuzzy, it was demonstrated that it has more preferable control accuracy.However, in above-mentioned document, The domain of fuzzy control is the quantizing factor by manual on-line tuning predicated error and predicated error variable quantity and obtains, because This, exact value of the controlled quentity controlled variable in domain is difficult timely obtain.

The content of the invention

It is an object of the invention to provide a kind of decomposition furnace outlet temperature sliding-mode control based on regression model, it is intended to The fuzzy control domain for solving existing decomposition furnace outlet temperature sliding-mode control presence is predicted by manual on-line tuning The quantizing factor of error and predicated error variable quantity and obtain, exact value of the controlled quentity controlled variable in domain be difficult it is timely obtain, cause The problem of precision of method is relatively low.

The present invention is achieved in that a kind of decomposition furnace outlet temperature sliding-mode control based on regression model, the base Comprise the following steps in the decomposition furnace outlet temperature sliding-mode control of regression model：

Step one, according to the relation between decomposition furnace outlet temperature and hello coal amount, using algorithm with regress analysis method, set up and decompose Mathematical modeling of the stove under typical condition；

Dore furnace mathematical modeling is：

In formula, y (n+1) is n+1 decomposition furnace outlet temperature, and y (n) is n decomposition furnace outlet temperature, and y (n-1) decomposes for n-1 Heater outlet temperature, α₁, α₂, α₃, α₄For model parameter to be identified, u (n) is to feed coal amount at the n moment；

Step 2, according to dore furnace mathematical modeling, asks for optimum control amount using adaptive tendency rate, sets up dore furnace Sliding mode controller, so as to realize decomposition furnace outlet temperature control；

Adaptively tendency rate is：

In formula, q=0.1, T=0.01, s (k), s (k+1) is the switching function at k and k+1 moment：

S (k)=C_eE=C_e(R (k)-x (k)),

Wherein, C_e=[10,1], R (k)=[r (k), dr (k)], r (k) is position command, and dr (k) is r (k) differential, x (k) it is decomposition furnace outlet temperature initial value；

Control rate is：

In formula, R (k+1) is the k+1 moment Position command, sgn () be sign function.

Further, need to draw decomposition furnace outlet temperature before step one and feed the relation between coal amount, and selection point 840 DEG C of heater outlet temperature~860 DEG C of operating modes as dore furnace are solved, it is 13.06t~15.1t to feed coal amount excursion.

Further, the relation between decomposition furnace outlet temperature and hello coal amount：When dore furnace is in normal operating conditions, point Solution heater outlet temperature is raised with the increase for feeding coal amount.

Further, the technological process that the input variable of decomposition furnace outlet temperature mathematical modeling is chosen in step one is as follows：

Step one, raw material are fed the connecting pipe of C1~C2 grades of cyclone cylinders by elevator；

Step 2, raw material bring C1 grades of cyclone cylinders into by the hot blast from C2 grades of cyclone cylinders and carry out gas-solid heat exchange, then by C1 The air valve discharge of level cyclone cylinder bottom, C2 grades of cyclone cylinders are brought into the connecting pipe of C2~C3 grades of cyclone cylinders, and by air-flow Interior continuation gas-solid heat exchange, so repeatedly；

Step 3, the material after preheating enters dore furnace via C4 grades of cyclone cylinder taperings, and coal dust is then given in the middle part of dore furnace Coal mouthful enters dore furnace；

Step 4, gooseneck of the material through dore furnace top come out from dore furnace enters C5 grades of cyclone cylinders；Finally, by C5 Calcined into rotary kiln in the tapering of level cyclone cylinder.

Further, the method that dore furnace mathematical modeling is obtained includes：

It is typical condition to work in 840 DEG C -860 DEG C according to decomposition furnace outlet temperature, i.e., decomposition furnace outlet temperature is at 840 DEG C During to 860 DEG C, dore furnace operating efficiency highest, using algorithm with regress analysis method to decomposition furnace outlet temperature in 840 DEG C of -860 DEG C of areas It is interior to model；

Decomposition furnace outlet temperature is linear with the inner link for feeding coal amount sampled value, i.e., by the measurement apparatus of control system Decomposition furnace outlet gas temperature and hello coal amount are sampled, n group measurement data is obtained, structure type is as follows：

In formula, y₀, y₂..., y_n+1For t₀To t_n+1The decomposition furnace outlet temperature at moment, u₁, u₂... u_nFor t₁To t_nMoment Feed coal amount, α₁, α₂, α₃, α₄For model parameter to be identified：

(1) formula is rewritten into matrix form, obtained：

Y=A α (2)

In formula：

Y=[y (2), y (3) ..., y (n+1)]^T

α=[α₁, α₂, α₃, α₄]^T

According to least-squares estimation, obtaining regression equation is：

Regression coefficient is：

α=(A^TA)^-1A^Ty (4)

Thus, the decomposition furnace outlet temperature mathematical modeling based on least square learning algorithm is：

Further, the acquisition methods of adaptive tendency rate and control rate formula：

According to the mathematical modeling set up, by formulaObtain following state Equation：

A(z^-1) y (k+1)=B (z^-1)u(k)+θ (7)

Wherein：

A(z^-1)=1-0.0045z^-1-0.0002z^-2

B(z^-1)=9.6468

θ=718.9233

Formula (7) is rewritten into following form：

X (k+1)=Ax (k)+Bu (k)+θ (8)

Y (k)=Cx (k)

Wherein：

C=[1 0]；

Switching function is chosen for：

S (k)=C_eE=C_e(R(k)-x(k)) (9)

Wherein, C_e=[c, 1]；R=[r (k), dr (k)], r (k) are position command, and formula (9) is rewritten into following form：

S (k+1)=C_e(R(k+1)-x(k+1))

=C_e(R(k+1)-Ax(k)-Bu(k)-θ) (10)

=C_eR(k+1)-C_eAx(k)-C_eBu(k)-C_eθ

Control rate is：

U (k)=(C_eB)^-1(C_eR(k+1)-C_eAx(k)-C_eθ-s(k+1)) (11)

Discrete tendency rate based on index is：

S (k+1)=s (k)+T (- ε sgn (s (k))-qs (k)) (12)

Wherein, ε ＞ 0；Q ＞ 0；1-qT ＞ 0；T is sampling time, T<<1.0, according to (12), formula (11) is rewritten into：

U (k)=(C_eB)^-1(C_eR(k+1)-C_eAx(k)-C_eθ-s(k)-ds(k)) (13)

Wherein, ds (k)=- ε T sgn (s (k))-qTs (k)；

According to formula (12), obtain：

There are following three kinds of situations by formula (14)：

WhenHave：

Therefore, | p | ＜ 1；| s (k+1) | ＜ | s (k) |；| s (k) | successively decrease；

WhenHave：

Therefore, | p | ＞ 1；| s (k+1) | ＞ | s (k) |；| s (k) | it is incremented by；

WhenHave：

Therefore, | p |=1；| s (k+1) |=| s (k) |；| s (k) | it is concussion；

Obviously, | s (k) | need to meet：

Formula (15) is rewritten into：

Take ε=| s (k) |/2,Formula (12) is rewritten into：

Adaptive tendency rate：

Control rate is：

The decomposition furnace outlet temperature sliding-mode control based on regression model that the present invention is provided, sets up adaptive sliding mode control Device processed, with stronger robustness and consistency, can accurately realize the control of decomposition furnace outlet temperature, specific advantage is as follows：By Fig. 4-5 understands that institute's established model can accurately reflect decomposition furnace outlet temperature change state, and error meets industry within ± 1 DEG C Production requirement；From Fig. 6-8, designed sliding mode controller, which can be adjusted accurately, feeds coal amount to realize the outlet temperature of dore furnace The control of degree, regulating time is within 1 minute, steady-state error back to zero, precise control；As shown in Figure 9, adaptive rate ε is in control Initial stage is larger, increases over time, and its value is gradually reduced back to zero, has both avoided the buffeting of system, and system arrival is improved again The speed of diverter surface.Therefore, optimal control of the present invention to realize industry spot dore furnace provides new approaches；

Compared with prior art, beneficial effects of the present invention are：

1st, mathematical modeling of the cement decomposing furnace outlet temperature at 840 DEG C -860 DEG C, the temperature model are established in the present invention The typical condition for dore furnace is enclosed, institute's established model can fully reflect temperature variations of the dore furnace under typical condition, had There is certain specific aim.

2nd, dimension difference between sample data is considered in the present invention, employs sample data normalization, can preferably dig Excavate the relation between data.

3rd, the structure of controller employs sliding mode variable structure control method, fast response time, physics realization letter in the present invention It is single.

4th, optimum control amount is asked for using adaptive exponential approach rate in the present invention, system mode is not shaken, progressive Tend to be balanced a little zero, control input signal smoothing is without buffeting.

Brief description of the drawings

Fig. 1 is the decomposition furnace outlet temperature sliding-mode control flow provided in an embodiment of the present invention based on regression model Figure；

Fig. 2 is calciner temperature control principle drawing provided in an embodiment of the present invention；

Fig. 3 is cement predecomposition process chart provided in an embodiment of the present invention；

Fig. 4 is decomposition furnace outlet temperature Model fitting curve map provided in an embodiment of the present invention；

Fig. 5 is decomposition furnace outlet temperature error of mathematical model curve map provided in an embodiment of the present invention；

Fig. 6 is decomposition furnace outlet temperature control curve figure provided in an embodiment of the present invention；

Fig. 7 is decomposition furnace outlet temperature tracking error curve map provided in an embodiment of the present invention；

Fig. 8 is controlled quentity controlled variable provided in an embodiment of the present invention (feeding coal) change curve；

It is adaptive rate ε change curves provided in an embodiment of the present invention that Fig. 9, which is,.

Embodiment

In order to make the purpose , technical scheme and advantage of the present invention be clearer, with reference to embodiments, to the present invention It is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not used to Limit the present invention.

Below in conjunction with the accompanying drawings and specific embodiment to the present invention application principle be further described.

Controller of Temperature of Cement Decomposing Furnace schematic diagram is as shown in Figure 2；

As shown in figure 1, the decomposition furnace outlet temperature sliding-mode control based on regression model of the embodiment of the present invention includes Following steps：

S101：According to cement predecomposition technological process and historical data, draw between decomposition furnace outlet temperature and hello coal amount Relation, and choose 840 DEG C of decomposition furnace outlet temperature~860 DEG C of typical conditions as dore furnace；

S102：Relation according to step S101, using algorithm with regress analysis method, sets up dore furnace under typical condition Mathematical modeling；

S103：Mathematical modeling according to step S102, optimum control amount is asked for using adaptive tendency rate, is set up The sliding mode controller of dore furnace, so as to realize decomposition furnace outlet temperature control.

Each step is described in further detail below：

In step S101：Relation between cement predecomposition technological process and decomposition furnace outlet temperature and hello coal amount；

The input variable of decomposition furnace outlet temperature mathematical modeling is chosen according to cement predecomposition technological process in the present invention , its process chart is as shown in figure 3, particularly may be divided into following steps：

Step one, raw material are fed the connecting pipe of C1~C2 grades of cyclone cylinders by elevator；

Step 2, raw material bring C1 grades of cyclone cylinders into by the hot blast from C2 grades of cyclone cylinders and carry out gas-solid heat exchange, then by C1 The air valve discharge of level cyclone cylinder bottom, C2 grades of cyclone cylinders are brought into the connecting pipe of C2~C3 grades of cyclone cylinders, and by air-flow Interior continuation gas-solid heat exchange, so repeatedly；

Step 3, the material after preheating enters dore furnace via C4 grades of cyclone cylinder taperings, and coal dust is then given in the middle part of dore furnace Coal mouthful enters dore furnace；Fully mix due to pulverized coal particle very little and with material, therefore the coal dust in dore furnace is with nonflame shape State is burnt；The heat that coal dust firing is discharged causes carbonate to absorb heat and reaction of decomposing by carbonate absorption；

Step 4, gooseneck of the material through dore furnace top come out from dore furnace enters C5 grades of cyclone cylinders；Finally, by C5 Calcined into rotary kiln in the tapering of level cyclone cylinder；

Analyze the technical process to understand, during the predecomposition of dore furnace, because there are many reactions, cause influence The variable of calciner temperature change is various, and the analysis according to decomposition furnace structure, production technology and internal-combustion mechanism is can be found that In the case where the factors such as climatic environment, the hardness of raw material, granular size are certain, decomposition furnace outlet temperature by raw material discharge quantity, Tertiary air quantity and the influence for feeding coal amount；Further according to the experience of site operation personnel, in dore furnace normal work, tertiary air valve Aperture generally remains constant；Furthermore, due to raw material feed opening away from dore furnace farther out, by regulation feeding change calciner temperature tool There is larger delay, therefore, site operation personnel usually relies on regulation and feeds coal amount to change calciner temperature；Increase when feeding coal amount When, decomposition furnace outlet temperature is also accordingly raised, therefore chooses hello coal amount as the input variable of model, sets up decomposition furnace outlet temperature Spend single-input single-output mathematical modeling；

In step S102：Dore furnace is in the mathematical modeling under typical condition：

According to the data analysis in step S101, decomposition furnace outlet temperature can be obtained and the relation such as table 1 between coal amount is fed It is shown：

The decomposition furnace outlet temperature of table 1 and feed relation between coal amount and raw material feeding capacity

According to Field Force's working experience, it is its typical condition that decomposition furnace outlet temperature, which works in 840 DEG C -860 DEG C, that is, is divided Heater outlet temperature is solved at 840 DEG C to 860 DEG C, dore furnace operating efficiency highest；Therefore, using algorithm with regress analysis method to dore furnace Outlet temperature is modeled in 840 DEG C of -860 DEG C of intervals；

Due to decomposition furnace outlet temperature, excursion is very big (820 DEG C to 880 DEG C) in one day, however, selected builds The mould interval temperature difference is within 20 DEG C, it can be assumed that decomposition furnace outlet temperature is linear with the inner link for feeding coal amount sampled value 's；Decomposition furnace outlet gas temperature and hello coal amount are sampled by the measurement apparatus of control system, n groups measurement number is obtained According to its structure type is as follows：

In formula, y₀, y₂..., y_n+1For t₀To t_n+1The decomposition furnace outlet temperature at moment, u₁, u₂... u_nFor t₁To t_nMoment Feed coal amount, α₁, α₂, α₃, α₄For model parameter to be identified；

(1) formula is rewritten into matrix form, obtained

Y=A α (2)

In formula：

Y=[y (2), y (3) ..., y (n+1)]^T

α=[α₁, α₂, α₃, α₄]^T

According to least-squares estimation, can obtain regression equation is：

Regression coefficient is：

α=(A^TA)^-1A^Ty (4)

It can thus be concluded that, the decomposition furnace outlet temperature mathematical modeling based on least square learning algorithm is：

Kernel program is write as follows：

clear all

Clc % cls

Data in the entitled 840-860 of % reading files tables of data

Caiyangzhi=xlsread (' C：\Users\asus\Desktop\840-860.xls′)；

Ceshishuju=xlsread (' C：\Users\asus\Desktop\840-860.xls′)；

a(1：200,1)=caiyangzhi (1：200,1)；

b(1：200,1)=caiyangzhi (1：200,2)；

c(1：200,1)=caiyangzhi (1：200,3)；

d(1：200,1)=caiyangzhi (1：200,4)；

e(1：200,1)=caiyangzhi (1：200,5)；

% asks for the maximum and minimum value of each variable

Minb=min (b)；Maxb=max (b)；

Minc=min (c)；Maxc=max (c)；

Mind=min (d)；Maxd=max (d)；

Mine=min (e)；Maxe=max (e)；

The processing of % data normalizations

For i=1：1：200

B (i, 1)=(b (i, 1)-minb)/(maxb-minb)；

C (i, 1)=(c (i, 1)-minc)/(maxc-minc)；

D (i, 1)=(d (i, 1)-mind)/(maxd-mind)；

E (i, 1)=(e (i, 1)-mine)/(maxe-mine)；

end

A1(1：200,1)=a；

A1(1：200,2)=b；

A1(1：200,3)=c；

A1(1：200,4)=d；

Y1(1：200,1)=e；

[b1, bint1, r1, rint1, stats1]=regress (Y1, A1,0.05)；

% takes in sampling data table preceding 100 row data as test data

Q2=caiyangzhi (1：100,5)；

Q1=A1 (1：100,1：4)*b1；

% data renormalizations

For k=1：1：100

Q1 (k, 1)=Q 1 (k, 1) * (maxe-mine)+mine；

End

% draws

figure(1)

X=1：100；

Plot (x, Q 1, ' r-o ', x, Q2, ' b--+ ')

figure(2)

X=1：100；

Plot (Q1-Q2, ' r ')；

Primitive modeling data are as shown in table 2 below：

The primitive modeling data of table 2

In summary, the mathematical modeling of decomposition furnace outlet temperature is：

In formula,

In step S103：The sliding mode controller of decomposition furnace outlet temperature：

According to the mathematical modeling set up in step S102, following state equation can be obtained by formula (6)：

A(z^-1) y (k+1)=B (z^-1)u(k)+θ (7)

Wherein,

A(z^-1)=1-0.0045z^-1-0.0002z^-2

B(z^-1)=9.6468

θ=718.9233

Formula (7) is rewritten into following form：

X (k+1)=Ax (k)+Bu (k)+θ (8)

Y (k)=Cx (k)

Wherein：

C=[1 0]；

Switching function is chosen for：

S (k)=C_eE=C_e(R(k)-x(k)) (9)

Wherein, C_e=[c, 1]；R=[r (k), dr (k)], r (k) are position command；Formula (9) can be rewritten into following shape Formula：

S (k+1)=C_e(R(k+1)-x(k+1))

=C_e(R(k+1)-Ax(k)-Bu(k)-θ) (10)

=C_eR(k+1)-C_eAx(k)-C_eBu(k)-C_eθ

Control rate is：

U (k)=(C_eB)^-1(C_eR(k+1)-C_eAxk)-C_eθ-s(k+1)) (11)

Discrete tendency rate based on index is：

S (k+1)=s (k)+T (- ε sgn (s (k))-qs (k)) (12)

Wherein, ε ＞ 0；Q ＞ 0；1-qT ＞ 0；T is sampling time, T<<1.0；According to (12), (11) it is rewritable into：

U (k)=(C_eB)^-1(C_eR(k+1)-C_eAx(k)-C_eθ-s(k)-ds(k)) (13)

Wherein, ds (k)=- ε T sgn (s (k))-qTs (k)；

According to formula (12), it can obtain：

There are following three kinds of situations by formula (14)：

1st, whenHave：

Therefore, | p | ＜ 1；| s (k+1) | ＜ | s (k) |；| s (k) | successively decrease；

2nd, whenHave：

Therefore, | p | ＞ 1；| s (k+1) | ＞ | s (k) |；| s (k) | it is incremented by；

3rd, whenHave：

Therefore, | p |=1；| s (k+1) |=| s (k) |；| s (k) | it is concussion；

Obviously, | s (k) | need to meet：

Formula (15) can be rewritten into：

Take ε=| s (k) |/2,Formula (12) can be rewritten into：

Control rate is：

Stability analysis：

It can be obtained by formula (14)：

Therefore, sliding formwork approaching (17) meets the existence and reaching condition of sliding mode, and designed controller is steady Fixed；

Kernel program is write as follows：

% cls

clear all；

close all；

% sets sampling time and parameter

Ts=0.01；

A=[- 0.0002, -0.0045；1,0]；

B=[9.6468；0]；

C=[1,0]；

D=[718.9233；0]；

% sets initial value

X=[840；840]；

R_1=855；R_2=850；

C=10；Q=0.1；

Ce=[c, 0]；

For k=1：1：8000

Time (k)=k*ts；

%, which is set, expects output temperature

R (k)=855.0；

% extrapolations

Dr (k)=(r (k)-r_1)/ts；

Dr_1=(r_1-r_2)/ts；

R1 (k)=2*r (k)-r_1；

Dr1 (k)=2*dr (k)-dr_1；

R=[r (k)；dr(k)]；

R1=[r1 (k)；dr1(k)]；

E=R-x；

E (k)=E (1)；

De (k)=E (2)；

S (k)=Ce*E；

The adaptive ε setting values of %

Eq (k)=abs (s (k))/2；

Ds (k)=- eq (k) * ts*sign (s (k))-q*ts*s (k)；

U (k)=inv (Ce*B) * (Ce*R1-Ce*A*x-Ce*D-s (k)-ds (k))；

Xx=x (1)；

X=A*x+B* (u (k))+D；

Y (k)=x (1)；

X (2)=xx；

% parameters update

R_2=r_1；

R_1=r (k)；

end

% draws

% controlling curves

figure(1)

Plot (time, r, ' r ', time, y, ' b ', ' linewidth ', 2)；

xlabel(′Time(s)′)；ylabel(′Outlet temperature(℃)′)；

% controls error

figure(2)

Plot (time, r-y)；

xlabel(′Time(s)′)；ylabel(′Tracking error(℃)′)；

% controlled quentity controlled variables (feed coal amount)

figure(3)

Plot (time, u, ' r ', ' linewidth ', 2)；

xlabel(′Time(s)′)；ylabel(′u′)；

% adaptive rates change

figure(4)；

Plot (time, eq, ' r ', ' linewidth ', 2)；

xlabel(′time(s)′)；ylabel(′adaptive\epsilon′)；

The specific embodiment of the present invention is described below；

Using Shandong cement plant production line data as foundation, the decomposition furnace outlet temperature mathematical modulo based on regression analysis is set up Type, separately takes 100 groups of data to be tested, the reliability of checking institute established model；Herein on basis, corresponding sliding formwork control is set up Device, sets controller parameter C_e=[10,1]；Q=0.1；T=0.01；

Fig. 4 shows the match value comparing result of decomposition furnace outlet temperature actual value and institute's established model output；Fig. 5 is corresponding Modeling error；Fig. 6 is decomposition furnace outlet temperature control curve figure；Fig. 7 is decomposition furnace outlet temperature tracking error curve map；Figure 8 be controlled quentity controlled variable (feeding coal) change curve；Fig. 9 is adaptive rate ε change curves.

From Fig. 4-5, institute's established model can accurately reflect decomposition furnace outlet temperature change state, error ± 1 DEG C with It is interior, meet demand of industrial production；From Fig. 6-8, designed sliding mode controller, which can be adjusted accurately, feeds coal amount to realize point The control of the outlet temperature of stove is solved, regulating time is within 1 minute, steady-state error back to zero, precise control；As shown in Figure 9, it is adaptive Should rate ε control initial stage it is larger, increase over time, its value is gradually reduced back to zero, both avoided the buffeting of system, again improve System reaches the speed of diverter surface..

The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all essences in the present invention Any modifications, equivalent substitutions and improvements made within refreshing and principle etc., should be included in the scope of the protection.

Claims (6)

1. a kind of decomposition furnace outlet temperature sliding-mode control based on regression model, it is characterised in that regression model should be based on Decomposition furnace outlet temperature sliding-mode control comprise the following steps：

Step one, according to the relation between decomposition furnace outlet temperature and hello coal amount, using algorithm with regress analysis method, set up dore furnace and exist Mathematical modeling under typical condition；

Dore furnace mathematical modeling is：

$y(n+1)={\widehat{\mathrm{\&alpha;}}}_1+{\widehat{\mathrm{\&alpha;}}}_{2}y\left(n\right)+{\widehat{\mathrm{\&alpha;}}}_{3}y(n-1)+{\widehat{\mathrm{\&alpha;}}}_{4}u\left(n\right)$

In formula, y (n+1) is the (n+1)th moment decomposition furnace outlet temperature, and y (n) is the n-th moment decomposition furnace outlet temperature, and y (n-1) is (n-1)th moment decomposition furnace outlet temperature,For model parameter to be identified, u (n) is to feed coal amount at the n moment；

Step 2, according to dore furnace mathematical modeling, optimum control amount is asked for using adaptive Reaching Law, the sliding formwork of dore furnace is set up Controller, so as to realize decomposition furnace outlet temperature control；

Adaptively Reaching Law is：

$s(k+1)-s\left(k\right)=-qTs\left(k\right)-\frac{s\left(k\right)}{2}Tsgn\left(s\right(k\left)\right)$

In formula, q=0.1, T=0.01, s (k), s (k+1) is the switching function at k and k+1 moment：

S (k)=C_eE=C_e(R(k)-x(k))

Wherein, C_e=[10,1], R (k)=[r (k), dr (k)], r (k) is position command, and dr (k) is r (k) differential, and x (k) is Decomposition furnace outlet temperature initial value；

Control law is：

$u\left(k\right)={\left(C_{e}B\right)}^{-1}\left(C_{e}R\right(k+1)-C_{e}Ax(k)-C_e\mathrm{\&theta;}-s(k)+qTs(k)+\frac{s\left(k\right)}{2}T\mathrm{sgn}(s\left(k\right)\left)\right)$

In formula,R (k+1) is the k+1 moment Position command matrix, sgn () is sign function.

2. the decomposition furnace outlet temperature sliding-mode control as claimed in claim 1 based on regression model, it is characterised in that Need to draw decomposition furnace outlet temperature before step one and feed the relation between coal amount, and choose 840 DEG C of decomposition furnace outlet temperature ~860 DEG C of operating modes as dore furnace, it is 13.06t~15.1t to feed coal amount excursion.

3. the decomposition furnace outlet temperature sliding-mode control as claimed in claim 1 based on regression model, it is characterised in that point Solve heater outlet temperature and feed the relation between coal amount：When dore furnace be in normal operating conditions when, decomposition furnace outlet temperature with feed The increase of coal amount and raise.

4. the decomposition furnace outlet temperature sliding-mode control as claimed in claim 1 based on regression model, it is characterised in that The technological process that the input variable of decomposition furnace outlet temperature mathematical modeling is chosen in step one is as follows：

Step 1.1, raw material are fed the connecting pipe of C1~C2 grades of cyclone cylinders by elevator；

Step 1.2, raw material bring C1 grades of cyclone cylinders into by the hot blast from C2 grades of cyclone cylinders and carry out gas-solid heat exchange, then by C1 grades of rotations Air duct bottom air valve discharge, into the connecting pipe of C2~C3 grades of cyclone cylinders, but by air-flow bring into C2 grades of cyclone cylinders after Continuous gas-solid heat exchange, so repeatedly；

Step 1.3, the material after preheating enters dore furnace via C4 grades of cyclone cylinder taperings, and coal dust is then from coal feeding hole in the middle part of dore furnace Into dore furnace；

Step 1.4, gooseneck of the material through dore furnace top come out from dore furnace enters C5 grades of cyclone cylinders；Finally, by C5 grades Calcined into rotary kiln in the tapering of cyclone cylinder.

5. the decomposition furnace outlet temperature sliding-mode control as claimed in claim 1 based on regression model, it is characterised in that point The method that solution stove mathematical modeling is obtained includes：

According to decomposition furnace outlet temperature work in 840 DEG C -860 DEG C for typical condition, i.e. decomposition furnace outlet temperature at 840 DEG C extremely At 860 DEG C, dore furnace operating efficiency highest, using algorithm with regress analysis method to decomposition furnace outlet temperature in 840 DEG C of -860 DEG C of intervals Interior modeling；

Decomposition furnace outlet temperature is linear with the inner link for feeding coal amount sampled value, i.e., by the measurement apparatus of control system to dividing Solution outlet of still gas temperature and hello coal amount are sampled, and obtain n group measurement data, structure type is as follows：

$\left\{\begin{array}{c}y\left(2\right)={\mathrm{\&alpha;}}_1+{\mathrm{\&alpha;}}_{2}y\left(1\right)+{\mathrm{\&alpha;}}_{3}y\left(0\right)+{\mathrm{\&alpha;}}_{4}u\left(1\right)\\ y\left(3\right)={\mathrm{\&alpha;}}_1+{\mathrm{\&alpha;}}_{2}y\left(2\right)+{\mathrm{\&alpha;}}_{3}y\left(1\right)+{\mathrm{\&alpha;}}_{4}u\left(2\right)\\ \mathrm{......}\\ y\left(n\right)={\mathrm{\&alpha;}}_1+{\mathrm{\&alpha;}}_{2}y(n-1)+{\mathrm{\&alpha;}}_{3}y(n-2)+{\mathrm{\&alpha;}}_{4}u(n-1)\\ y(n+1)={\mathrm{\&alpha;}}_1+{\mathrm{\&alpha;}}_{2}y\left(n\right)+{\mathrm{\&alpha;}}_{3}y(n-1)+{\mathrm{\&alpha;}}_{4}u\left(n\right)\end{array}\right.---\left(1\right)$

In formula, y₀To feed decomposition furnace outlet temperature when coal amount is 0, y₁,y₂,…,y_n+1For t₁To t_n+1The dore furnace at moment goes out Mouth temperature, u₁,u₂,…u_nFor t₁To t_nHello the coal amount at moment, α₁,α₂,α₃,α₄For model regression parameter to be identified：

(1) formula is rewritten into matrix form, obtained：

Y=A α (2)

In formula：

Y=[y (2), y (3) ..., y (n+1)]^T

α=[α₁,α₂,α₃,α₄]^T

$A={\left[\begin{array}{cccc}1& y\left(1\right)& y\left(0\right)& u\left(1\right)\\ 1& y\left(1\right)& y\left(1\right)& u\left(2\right)\\ .& .& .& .\\ .& .& .& .\\ .& .& .& .\\ 1& y\left(n\right)& y(n-1)& u\left(n\right)\end{array}\right]}_{n\&times;4}$

According to least-squares estimation, obtaining regression equation is：

$y(n+1)={\mathrm{\&alpha;}}_1+{\mathrm{\&alpha;}}_{2}y\left(n\right)+{\mathrm{\&alpha;}}_{3}y(n-1)+{\mathrm{\&alpha;}}_{4}u\left(n\right)---\left(3\right)$

Regression coefficient is：

α=(A^TA)^-1A^Ty (4)

Thus, the decomposition furnace outlet temperature mathematical modeling based on least square learning algorithm is：

$y(n+1)={\widehat{\mathrm{\&alpha;}}}_1+{\widehat{\mathrm{\&alpha;}}}_{2}y\left(n\right)+{\widehat{\mathrm{\&alpha;}}}_{3}y(n-1)+{\widehat{\mathrm{\&alpha;}}}_{4}u\left(n\right)$

Wherein, y (n+1) withRegression relation, α each other₁,α₂,α₃,α₄Respectively withRegression relation each other.

6. the decomposition furnace outlet temperature sliding-mode control as claimed in claim 1 based on regression model, it is characterised in that from Adapt to the acquisition methods of Reaching Law and control law formula：

According to the mathematical modeling set up, by formulaObtain following state side Journey：

A(z^-1) y (k+1)=B (z^-1)u(k)+θ (7)

Wherein：

A(z^-1)=1-0.0045z^-1-0.0002z^-2

B(z^-1)=9.6468

θ=718.9233

Formula (7) is rewritten into following form：

$\begin{array}{c}x(k+1)=Ax\left(k\right)+Bu\left(k\right)+\mathrm{\&theta;}\\ y\left(k\right)=Cx\left(k\right)\end{array}---\left(8\right)$

Wherein：

$A=\left[\begin{array}{cc}-0.0045& -0.0002\\ 1& 0\end{array}\right];B=\left[\begin{array}{c}9.6468\\ 0\end{array}\right];C=\&lsqb;10\&rsqb;;\mathrm{\&theta;}=\left[\begin{array}{c}718.9233\\ 0\end{array}\right]$

Switching function is chosen for：

S (k)=C_eE=C_e(R(k)-x(k)) (9)

Wherein, C_e=[c, 1]；R=[r (k), dr (k)], r (k) are position command, and formula (9) is rewritten into following form：

$\begin{array}{c}s(k+1)=C_e\left(R\right(k+1)-x(k+1\left)\right)\\ =C_e\left(R\right(k+1)-Ax(k)-Bu(k)-\mathrm{\&theta;})\\ =C_{e}R(k+1)-C_{e}Ax\left(k\right)-C_{e}Bu\left(k\right)-C_e\mathrm{\&theta;}\end{array}---\left(10\right)$

Control law is：

U (k)=(C_eB)^-1(C_eR(k+1)-C_eAx(k)-C_eθ-s(k+1)) (11)

Discrete reaching law based on index is：

S (k+1)=s (k)+T (- ε sgn (s (k))-qs (k)) (12)

Wherein, ε ＞ 0；Q ＞ 0；1-qT ＞ 0；T is sampling time, T<<1.0, according to (12), formula (11) is rewritten into：

U (k)=(C_eB)^-1(C_eR(k+1)-C_eAx(k)-C_eθ-s(k)-ds(k)) (13)

Wherein, ds (k)=- ε Tsgn (s (k))-qTs (k)；

According to formula (12), obtain：

$\begin{array}{c}s(k+1)=(1-qT)s\left(k\right)-\mathrm{\&epsiv;}T\frac{s\left(k\right)}{\left|s\left(k\right)\right|}\\ =(1-qT-\frac{\mathrm{\&epsiv;}T}{\left|s\left(k\right)\right|})s\left(k\right)=ps\left(k\right)\end{array}---\left(14\right)$

There are following three kinds of situations by formula (14)：

WhenHave：

$p>1-qT-\frac{\mathrm{\&epsiv;}T(2-qT)}{\mathrm{\&epsiv;}T}=1-qT-(2-qT)=-1$

Therefore, | p | ＜ 1；| s (k+1) | ＜ | s (k) |；| s (k) | successively decrease；

WhenHave：

$p<1-qT-\frac{\mathrm{\&epsiv;}T(2-qT)}{\mathrm{\&epsiv;}T}=1-qT-(2-qT)=-1$

Therefore, | p | ＞ 1；| s (k+1) | ＞ | s (k) |；| s (k) | it is incremented by；

WhenHave：

$p=1-qT-\frac{\mathrm{\&epsiv;}T(2-qT)}{\mathrm{\&epsiv;}T}=1-qT-(2-qT)=-1$

Therefore, | p |=1；| s (k+1) |=| s (k) |；| s (k) | it is concussion；

Obviously, | s (k) | need to meet：

$\left|s\left(k\right)\right|>\frac{\mathrm{\&epsiv;}T}{2-qT}---\left(15\right)$

Formula (15) is rewritten into：

$\mathrm{\&epsiv;}<\frac{1}{T}(2-qT)\left|s\left(k\right)\right|---\left(16\right)$

Take ε=| s (k) |/2,Formula (12) is rewritten into：

Adaptive Reaching Law：

Control law is：

$\begin{array}{c}u\left(k\right)={\left(C_{e}B\right)}^{-1}\left(C_{e}R\right(k+1)-C_{e}Ax(k)-C_e\mathrm{\&theta;}\\ -s\left(k\right)+qTs\left(k\right)+\frac{s\left(k\right)}{2}T\mathrm{sgn}\left(s\right(k\left)\right))\end{array}.$